Application of Fourier transform near-infrared spectroscopy to the quantification and monitoring of carbonyl value in frying oils

Abstract

A rapid and convenient method for determining carbonyl value (CV) in frying oil using Fourier transform near-infrared (FTNIR) spectroscopy was established. A total of 144 oil samples were subjected to CV determination using the 2,4-dinitrophenylhydrazine (2,4-DNPH) method and FTNIR spectroscopy combined with partial least square (PLS) modeling. The PLS model worked best with pretreated spectra in the wavelength range of 9702 cm⁻¹ to 4531 cm⁻¹, resulting in a root mean square error of calibration (RMSEC) of 1.47 mmol kg⁻¹ and prediction (RMSEP) of 1.73 mmol kg⁻¹. Studies related to the monitoring of CV changes in a simulated frying process of oil samples were then performed to evaluate the feasibility of the method for online monitoring of oil quality. Results showed that the model-predicted value was close to the true value determined by the 2,4-DNPH method (R = 0.989). Therefore, the FTNIR method can be used for quantifying and monitoring CV with the advantages of simple operation and no pollution.